1: %\documentclass[preprint]{aastex}

2: \documentclass[useAMS,usenatbib]{mn2e}

3: \bibliographystyle{mn2e}

4: \usepackage{epsfig}

5: \usepackage{amsmath}

6: %\usepackage{mn}

7: %\usepackage{mn-nat}

8: 

9: \newcommand{\be}{\begin{equation}}

10: \newcommand{\beq}{\begin{equation}}

11: \newcommand{\ba}{\begin{eqnarray}}

12: \newcommand{\ee}{\end{equation}}

13: \newcommand{\eeq}{\end{equation}}

14: \newcommand{\ea}{\end{eqnarray}}

15: \newcommand{\msun}{$M_{\odot}\hspace{1mm}$}

16: \newcommand{\wmap}{{\it WMAP }}

17: \newcommand{\kel}{$^{\circ} K\hspace{1mm}$}

18: \newcommand{\hs}{\hspace{1mm}}

19: \newcommand{\vs}{\vspace{2mm}}

20: \newcommand{\lya}{Ly$\alpha \hspace{1mm}$}

21: \newcommand{\ha}{H$\alpha \hspace{1mm}$}

22: \newcommand{\kms}{km s$^{-1}\hspace{1mm}$}

23: \newcommand{\Mpc}{{\rm Mpc}}

24: \newcommand{\dPsidL}{$\partial \log \Psi/\partial \log L_B$}

25: \newcommand{\apj}{ApJ}

26: \newcommand{\aap}{A\&A}

27: \newcommand{\apjl}{ApJL}

28: \newcommand{\mnras}{MNRAS}

29: \newcommand{\aj}{AJ}

30: \newcommand{\apjs}{ApJS}

31: \newcommand{\nat}{{\it Nature}}

32: \newcommand{\araa}{ARA\&A}

33: \newcommand{\pasj}{PASJ}

34: \newcommand{\llya}{$L_{{\rm Ly}\alpha}$}

35: \newcommand{\physrep}{Physics Reports}

36: % definition to produce a "less than or similar to" symbol

37: \def\lsim{~\rlap{$<$}{\lower 1.0ex\hbox{$\sim$}}}

38: 

39: % definition to produce a "greater than or similar to" symbol

40: \def\gsim{~\rlap{$>$}{\lower 1.0ex\hbox{$\sim$}}}

41: %

42: 

43: %\title[Ly$\alpha$ Radiation Pressure]{Dynamical Effects

44: %of Ly$\alpha$ Radiation Pressure around Galaxies}

45: \title[Ly$\alpha$ Radiation Pressure]{Ly$\alpha$ Driven Outflows Around Star Forming Galaxies}

46: 

47: \author[Mark Dijkstra \& Abraham Loeb]{Mark Dijkstra\thanks{E-mail:mdijkstr@cfa.harvard.edu} \& Abraham Loeb\thanks{E-mail:aloeb@cfa.harvard.edu}\\

48: Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138, USA}

49: 

50: \def\LaTeX{L\kern-.36em\raise.3ex\hbox{a}\kern-.15em

51:     T\kern-.1667em\lower.7ex\hbox{E}\kern-.125emX}

52: 

53: \newtheorem{theorem}{Theorem}[section]

54: 

55: %\voffset=-15mm

56: 

57: \begin{document}

58: 

59: \date{\today}

60: \pagerange{\pageref{firstpage}--\pageref{lastpage}} \pubyear{2007}

61: 

62: \maketitle

63: 

64: \label{firstpage}

65: \begin{abstract}

66: We present accurate Monte-Carlo calculations of Ly$\alpha$ radiation

67: pressure in a range of models which represent galaxies during various

68: epochs of our Universe. We show that the radiation force that Ly$\alpha$

69: photons exert on hydrogen gas in the neutral intergalactic medium (IGM),

70: that surrounds minihalos that host the first stars, may exceed gravity by

71: orders of magnitude and drive supersonic winds. Ly$\alpha$ radiation

72: pressure may also dominate over gravity in the neutral IGM that surrounds

73: the HII regions produced by the first galaxies. However, the radiation

74: force is likely too weak to result in supersonic outflows in this

75: case. Furthermore, we show that Ly$\alpha$ radiation pressure may drive

76: outflows in the interstellar medium of star forming galaxies that reach

77: hundreds of km s$^{-1}$. This mechanism could also operate at lower

78: redshifts $z\lsim 6$, and may have already been indirectly detected in the

79: spectral line shape of observed Ly$\alpha$ emission lines.  \end{abstract}

80: 

81: \begin{keywords}

82: cosmology: theory--galaxies: high redshift--radiation mechanisms: general--radiative transfer--ISM: bubbles 

83: \end{keywords}

84:  

85: \section{Introduction}

86: \label{sec:intro}

87: 

88: HII regions around massive stars convert a significant fraction of the

89: total bolometric luminosity of young galaxies into Ly$\alpha$ line emission

90: \citep{PP67,S03}. This Ly$\alpha$ radiation can exert a large force on

91: surrounding neutral gas, as the Ly$\alpha$ transition has a

92: cross-section that is $\sim 7$ orders of magnitude

93: larger than the Thomson cross-section, when averaged over a frequency band as wide as the resonance frequency itself \citep[e.g.][]{Loeb01}. Not

94: surprisingly, the impact of Ly$\alpha$ radiation pressure on the formation

95: of galaxies has been discussed extensively

96: \citep[e.g][]{Cox85,EF86,Bithell90,Haenelt95,Oh02,McKee07}, but the

97: intricacies of Ly$\alpha$ radiative transfer in 3D complicated an accurate

98: numerical treatment of its dynamical effect on the gas. Nevertheless, an

99: approximate estimate can be obtained from simple energy considerations as

100: shown below.

101: 

102: Consider a self-gravitating gas cloud of total (baryons + dark matter) mass $M$ and radius $R$ that

103: contains a central Ly$\alpha$ source. The gravitational binding energy of

104: the baryons inside the cloud, $E_B\sim \Omega_bGM^2/(\Omega_m R)$, can be compared to the total energy in the

105: Ly$\alpha$ radiation field inside the cloud, $E_{\alpha}=L_{\alpha}\times

106: t_{\rm trap}$. Here, $L_{\alpha}$ is the Ly$\alpha$ luminosity of the

107: central source (in erg s$^{-1}$), and $t_{\rm trap}$ is the typical

108: trapping time of Ly$\alpha$ photons in the cloud owing to scattering on

109: hydrogen atoms. The Ly$\alpha$ radiation pressure would unbind the baryonic gas

110: from the cloud if $E_{\alpha}> E_B$, i.e. $L_{\alpha}>\Omega_bGM^2/(\Omega_mRt_{\rm trap})$

111: \citep[e.g.][]{Cox85,Bithell90,Oh02}. In this approach, $t_{\rm trap}$ is

112: one of the key parameters in setting the Ly$\alpha$ radiation

113: pressure. Calculations by \citet{Adams75} imply that $t_{\rm trap}\sim 15

114: t_{\rm light}$ for $3\lsim \log \tau_0 \lsim 5.5$, and $t_{\rm trap}\sim

115: 15(\tau_0/10^{5.5})^{1/3} t_{\rm light}$ otherwise, for a static, uniform,

116: infinite slab of material \citep[also see Fig~1 of][]{Bo79}. Here $\tau_0$

117: is the line center optical depth from the center to the edge of the slab,

118: and $t_{\rm light}$ is the light crossing time if the medium were

119: transparent (i.e. $t_{\rm light}=R/c$ in the case of the cloud described

120: above). Note however, that the precise value of $t_{\rm trap}$ depends on

121: other factors including for example, the gas distribution (clumpiness and

122: geometry), the velocity distribution of the gas, and the dust content of

123: the cloud \citep{Bo79}.  The Ly$\alpha$ radiation

124: pressure becomes comparable to gravity when

125: \begin{equation}

126: L_{\alpha,{\rm 41}}=1.0\Big{(}\frac{M}{10^{8}M_{\odot}}\Big{)}^{4/3}\Big{(}\frac{16}{1+z}\Big{)}^2\Big{(}\frac{15t_{\rm light}}{t_{\rm trap}}\Big{)}, 

127: \label{eq:lcrit}

128: \end{equation} 

129: where $L_{\alpha}=L_{\alpha,{\rm 41}}\times 10^{41}$ erg s$^{-1}$, and

130: where we have substituted the virial radius of a galaxy

131: mass $M$, $R_{\rm vir}=0.97$

132: kpc $\times(M/10^8M_{\odot})^{1/3}$$(1+z/16)^{-1}$, for $R$ (Eq.~24 of Barkana \& Loeb

133: 2001). For comparison, a star forming galaxy can generate a Ly$\alpha$

134: luminosity of $L_{\alpha}=(10^{42}-10^{43})\times({\rm

135: SFR}/M_{\odot}\hs{\rm yr}^{-1})$ erg s$^{-1}$, where the precise conversion

136: factor depends on the gas metallicity and the stellar initial mass function

137: \citep[e.g.][]{S03}. Therefore, a star formation rate of merely SFR$\gsim

138: 0.01$--$0.1M_{\odot}$ yr$^{-1}$ is needed to generate a Ly$\alpha$

139: luminosity that is capable of unbinding gas from a halo of mass

140: $10^8-10^9M_{\odot}$.

141: 

142: Halos of $\la 10^9M_\odot$ are very common at $z\gsim 6$, and have a

143: sufficiently large reservoir of baryons to sustain the above-mentioned star

144: formation rates for a prolonged time. In this paper we provide a more

145: detailed investigation of the magnitude of Ly$\alpha$ radiation pressure in

146: the environment of high-redshift star forming galaxies.

147: In particular, we use a Ly$\alpha$ Monte-Carlo radiative transfer code

148: \citep{mc} to compute Ly$\alpha$ radiation pressure in a wider range of

149: models.

150: Our treatment of radiative transfer and our focus on the environment of

151: high-redshift star forming galaxies, distinguish this paper from previous

152: work. We will show that the radiation force exerted by Ly$\alpha$ photons

153: on neutral hydrogen gas can exceed the gravitational force that binds the gas

154: to its host galaxy by orders of magnitude, and may drive supersonic

155: outflows of neutral gas both in the intergalactic and the interstellar

156: medium.

157: 

158: The outline of this paper is as follows.

159: %

160: In \S~\ref{sec:code} we describe how Ly$\alpha$ radiation pressure is

161: computed in the Monte-Carlo radiative transfer code, and show the tests

162: that are performed to test the accuracy of the code.

163: %

164: In \S~\ref{sec:result}, we present our numerical results.  Finally,

165: \S~\ref{sec:conc} summarizes the implications of our work and our main

166: conclusions. The cosmological parameter values used throughout our

167: discussion are

168: $(\Omega_m,\Omega_{\Lambda},\Omega_b,h)=(0.27,0.73,0.042,0.70)$

169: \citep{Komatsu08}.

170: 

171: 

172: \section{Ly$\alpha$ Radiation Pressure}

173: \label{sec:code}

174: 

175: The force $F_{\rm rad}$ experienced by an atom in a direction ${\bf n}$ is

176: related to the flux through a plane normal to ${\bf n}$,

177: \begin{equation}

178: F_{\rm rad}=\frac{4\pi}{c}\int d\nu \hs \sigma({\nu}) H(\nu),

179: \label{eq:flux1}

180: \end{equation} 

181: where $\sigma(\nu)$ is the Ly$\alpha$ absorption cross-section at frequency

182: $\nu$. The specific flux is given by $H(\nu)=\frac{1}{2}\int d\mu\hs \mu

183: I(\mu,\nu)$, where $I(\nu,\mu)$ is the specific intensity of the radiation

184: field \citep[see, e.g. Eq. 1.113 in][]{RL79}, and $\mu={\bf n}\cdot {\bf

185: k}$ in which ${\bf k}$ denotes the propagation direction of the radiation

186: (i.e. $\mu=1$ for radiation propagating perpendicular to the plane).

187: 

188: The specific intensity obeys the radiative transfer equation, which reads

189: (in spherical coordinates)

190: \begin{equation}

191: \mu \frac{\partial I}{\partial r}+\frac{(1-\mu^2)}{r}\frac{\partial

192: I}{\partial \mu}=\chi_{\nu}(J-I)+S_{\nu}(r),

193: \label{eq:RT}

194: \end{equation}

195: where in this equation $\mu \equiv {\bf r}\cdot {\bf k}/|{\bf r}|$,

196: $J(\nu)=\int d\mu\hs I(\mu,\nu)$ denotes the mean intensity, and $S_{\nu}(r)$ the emission function for newly created photons at frequency $\nu$ and radius $r$ (in photons cm$^{-3}$ s$^{-1}$ sr$^{-1}$ Hz$^{-1}$, see e.g. Loeb \& Rybicki

197: 1999). Furthermore, $\chi_{\nu}=\frac{h_P\nu_{\alpha}}{4\pi}\frac{B_{21}}{\sqrt{\pi}\Delta \nu_{\alpha}}\big{(}3n_1-n_2\big{)}\phi(\nu)$ denotes the opacity at frequency $\nu$ \citep[e.g.][]{RL79}, where $h_{\rm p}$ is Planck's constant, $\nu_\alpha=2.46\times 10^{15}$ Hz is the Ly$\alpha$ frequency, $n_{1(2)}$ is the number density of hydrogen atoms in their electronic ground (first excited) state, $B_{21}$ is the Einstein-B coefficient of the $2\rightarrow 1$ transition, $\phi(\nu)$ is the line profile function (e.g. Rybicki \& Lightman 1979, their Eq. 1.79), and $\Delta \nu_{\alpha}=\frac{v_{\rm th}}{c}$. Here, $v_{th}$ is the thermal velocity of the hydrogen atoms in the gas, given by $v_{th}=\sqrt{2k_B T/m_p}$, where $k_B$ is the Boltzmann constant, $T$ the gas temperature, and $m_p$ the proton mass.

198: 

199: Under the assumption that $I(\nu,\mu)$ has only a weak dependence on

200: direction (which is reasonable given that Ly$\alpha$ radiation scatters

201: very frequently), $I(\nu,\mu)$ can be expressed as a first-order Taylor

202: expansion in $\mu$, i.e. $I(\nu,\mu)=a(\nu)+b(\nu)\mu$. In this so-called

203: ``Eddington approximation'', the expression for flux simplifies to

204: \citep[e.g.][ their Eq. 1.118]{RL79}

205: \begin{equation}

206: H(\nu)=\frac{1}{3}\frac{dJ(\nu)}{d\tau}=\frac{c}{12\pi}\frac{du(\nu)}{d\tau},

207: \label{eq:flux2}

208: \end{equation} where we have used the relation, $u=4\pi J/c$, in which $u(\nu)$ is the specific energy density in the radiation field at a frequency $\nu$. Furthermore, we have decoupled the gas' absorption and emission functions from the Ly$\alpha$ radiation field, and assumed that all neutral hydrogen atoms are in their electronic ground state (this assumption is justified in more detail in Appendix~\ref{app:n2}), i.e. $n_2=0$ and $n_1=n_H$. Under this assumption,  $d\tau=\chi_{\nu}dr=n_H\sigma(\nu)dr$ with $n_H$ being the number density of neutral hydrogen atoms. Substituting this expression back into Eq.~(\ref{eq:flux1}) yields

209: \begin{equation}

210: F_{\rm rad}=\frac{1}{3n_H}\frac{d}{dr}\int d\nu \hs

211: u(\nu)=\frac{1}{3n_H}\frac{dU}{dr},

212: \label{eq:force}

213: \end{equation} 

214: where we defined $U \equiv \int d\nu\hs u(\nu)$. Note that the

215: cross-section does not appear in the final expression for the radiation

216: force\footnote{The right-hand-side of Eq.~(\ref{eq:force}) is analogous to

217: the usual pressure gradient force in fluid-dynamics which is not dependent

218: on the scattering cross-section of the fluid particles.}.

219: 

220: \subsection{Implementation in Monte-Carlo Technique}

221: \label{sec:radpresmc}

222: 

223: In our Monte-Carlo simulation we sample the gas density and velocity fields

224: with $N_s=5000$ concentric spherical shells. The radius, thickness, and volume of

225: shell $j$ are denoted by $r_j$, $dr_j$, and $V_j$, respectively. We compute

226: the radiation force using two approaches:

227: 

228: \begin{itemize}

229: 

230: \item In the first approach, we calculate the energy density ($U$ in

231: Eq.~\ref{eq:force}) in the Ly$\alpha$ radiation field as a function of

232: radius: Using the Monte-Carlo simulation we compute the average time that

233: photons spend in shell $j$, which we denote by $\langle t \rangle_j$. The

234: total number of photons that is present in shell $j$ at any given time is

235: then given by $N_{\alpha,j}=\dot{N}_{\alpha}\times \langle t \rangle_j$,

236: where $\dot{N}_{\alpha}$ is the rate at which photons are emitted. This

237: yields the energy density, $U_j=N_{\alpha,j}h\nu_{\alpha}/V_j$. Finally, Eq.~(\ref{eq:force}) is used to compute the radiation force on atoms in shell $j$. Note that estimators of the energy density in -and the momentum transfer by- a radiation field in a more general context is discussed by e.g. \citet{Lucy99} and \citet{Lucy07}.

238: 

239: \item In the second approach, we calculate the momentum transfer from a

240: Ly$\alpha$ photon to an atom in each scattering event, $\Delta p=h_{\rm p}

241: ({\bf k}_{\rm in}-{\bf k}_{\rm out})/2\pi$. Here, ${\bf k}_{\rm in}$ and ${\bf k}_{\rm out}$ are the photons

242: wavevectors before and after scattering. We compute the average {\it total

243: momentum transfer} (i.e. summed over all scattering events) per photon in

244: shell $j$, $\langle \Delta \mathcal{P}\rangle_j$, and obtain the total

245: momentum transfer from $\dot{P}_{\alpha,j}=\dot{N}_{\alpha}\times \langle

246: \Delta \mathcal{P}\rangle_j$. The force on an individual atom is obtained

247: by dividing by the total number of hydrogen atoms in shell $j$, i.e

248: $F_j=\dot{P}_{\alpha,j}/(V_j\times n_{H,j})$.

249: 

250: \end{itemize}

251: 

252: Both methods should give identical results, provided that the Eddington

253: approximation holds.

254: 

255: \subsection{Test Case: Sources in a Neutral Comoving IGM}

256: \label{sec:test}

257: 

258: We begin by considering a Ly$\alpha$ point source at a redshift $z=10$

259: embedded in a neutral intergalactic medium (IGM) that is expanding with the

260: Hubble flow. The photons scatter and diffuse away from the source while

261: Hubble expansion redshifts the photons away from resonance. In this case, the angle-averaged intensity $J(\nu)$ and its radial dependence can be

262: calculated analytically \citep{LR99}. The availability of analytic

263: expressions for $J(\nu,r)$, and therefore the radiation force $F_{\rm rad}$

264: (through Eq.~\ref{eq:force}), makes this a good test case for our code.

265: 

266: In Figure~\ref{fig:uden} we plot the radial dependence of the energy

267: density (in erg cm$^{-3}$) in the Ly$\alpha$ radiation field for a

268: model in which the central source is emitting

269: $\dot{N}_{\alpha,54}\times 10^{54}$ photons s$^{-1}$ (where we have

270: introduced the dimensionless quantity $\dot{N}_{\alpha,54}\equiv

271: (\dot{N}_{\alpha}/10^{54}$ photons s$^{-1}$). This corresponds to a

272: luminosity of $L_{\alpha}=\dot{N}_{\alpha,54} \times 1.6\times

273: 10^{43}$ erg s$^{-1}$, which represents a bright Ly$\alpha$ emitting

274: galaxy \citep[e.g.][]{Ouchi08}. The {\it blue dotted line} shows the

275: energy density if the IGM were fully transparent to Ly$\alpha$

276: radiation. In this hypothetical case all photons stream radially

277: outward, and the energy density is given by $L_{\alpha}/(4\pi r^2

278: c)$. The {\it red dashed line} shows the energy density,

279: $U(r)=\frac{4\pi}{c}\int d\nu\hs J(\nu,r)$, derived from the analytic

280: expression for $J(r,\nu)$ given in Loeb \& Rybicki (1999, their

281: Eq.~21), while the {\it black histogram} shows the energy density

282: extracted from the simulation (\S~\ref{sec:radpresmc}). Clearly, the

283: analytic and Monte-Carlo calculations yield consistent

284: results. Scattering reduces the effective speed at which photons

285: propagate radially outward, which enhances their energy density

286: (especially at small radii) relative to the transparent case. At

287: sufficiently large distances however, the photons have redshifted far

288: enough from resonance that they are propagating almost freely to the

289: observer, and the energy density approaches $L_{\alpha}/4\pi r^2

290: c$. We note that at a sufficiently high value of

291: $\dot{N}_{\alpha,54}$, the fraction of hydrogen atoms that populate

292: the 2p (and 2s) levels is non-negligible and our assumption that (almost) all

293: of the atoms populate their electronic ground state becomes invalid

294: (so that the solution for $U(r)$ in Figure~\ref{fig:uden} breaks

295: down). However, as we show in Appendix~\ref{app:n2}, this only occurs

296: when $\dot{N}_{\alpha,54} \gsim 10^7$, well beyond the regime

297: considered in this paper.

298: 

299: \begin{figure}

300: \vbox{\centerline{\epsfig{file=fig1.eps,angle=270,width=8.0truecm}}}

301: \caption[]{Radial profile of the energy density $U(r)$ (erg cm$^{-3}$) in

302: the Ly$\alpha$ radiation field surrounding a central source that is

303: emitting $\dot{N}_{\alpha,54} \times 10^{54}$ photons s$^{-1}$ into an expanding neutral IGM. The {\it blue dotted line} shows $U(r)$ if the IGM were fully transparent. The {\it black solid

304: (red dashed) line} shows $U(r)$ when radiative transfer is included using an

305: analytic (Monte-Carlo) approach (see text). Scattering reduces the speed at

306: which Ly$\alpha$ photons are propagating radially outward, increasing

307: $U(r)$ relative to the transparent case.}

308: \label{fig:uden}

309: \end{figure}

310: \begin{figure}

311: \vbox{\centerline{\epsfig{file=fig2.eps,angle=270,width=8.0truecm}}}

312: \caption[]{The ratio of radiation to gravitational force on a hydrogen atom

313: as a function of physical radius in kpc. To scale out the dependence of

314: this ratio on halo mass, $M_h$, and production rate of Ly$\alpha$ photons

315: by the central source, $\dot{N}_{\alpha}$, the vertical axis is normalized

316: by $M_{h,11}/\dot{N}_{\alpha,54}$ (see text). The {\it black dotted (grey solid) line}

317: was obtained by applying Eq.~(\ref{eq:force}) to the energy density $U(r)$

318: that was obtained by using the analytic (Monte-Carlo) approach (see

319: Fig~\ref{fig:uden}). The {\it black solid line} was obtained by directly computing

320: the momentum transfer rate from photons to atoms in the Monte-Carlo code as

321: outlined in \S~\ref{sec:radpresmc}. The results demonstrate that ({\it i})

322: the radiation force exceeds gravity at $r<10

323: (\dot{N}_{\alpha,54}/M_{h,11})$ kpc, and ({\it ii}) both methods yield

324: consistent results.}

325: \label{fig:force1}

326: \end{figure}

327: %

328: In Figure~\ref{fig:force1} we compare the radiation force to the

329: gravitational force on a single hydrogen atom, $F_{\rm grav}=GM(<r)m_p/r^2$,

330: where $M(<r)$ is the total mass enclosed within a radius $r$). We plot the

331: ratio $F_{\rm rad}/F_{\rm grav}$ scaled by $M=10^{11}M_{\odot}$

332: \footnote{The number density of halos more massive than $10^{11}M_{\odot}$

333: at $z=10$ is $\sim 10^{-7}$ comoving Mpc$^{-3}$, implying that these rare

334: halos are among the most massive ones in existence at that early cosmic

335: time.}. The {\it black dotted line (grey solid histogram)} was calculated by applying

336: Eq.~(\ref{eq:force}) to the energy density $U(r)$ that was obtained by

337: using the analytic (Monte-Carlo) approach (also see

338: Fig~\ref{fig:uden}). For comparison, the {\it black solid histogram} was obtained by

339: directly computing the momentum transfer rate from photons to atoms as

340: outlined in \S~\ref{sec:radpresmc}. Figure~\ref{fig:force1} shows that the

341: radiation force overwhelms gravity at small radii. The energy density

342: scales approximately as $\partial \log U/\partial \log r \sim-2.3$

343: (Fig~\ref{fig:uden}). Therefore, $F_{\rm rad}/F_{\rm grav} \propto r^{-1.3}$ and

344: reaches unity at $r \sim 10$ physical kpc.

345: 

346: The radiation force increases linearly with $\dot{N}_{\alpha}$ while the

347: gravitational force scales as $M$. Thus, $F_{\rm rad}/F_{\rm grav}$ scales linearly

348: with the ratio $\mathcal{R}\equiv \dot{N}_{\alpha}/M$. To scale out the

349: dependence on $\mathcal{R}$, the vertical axis shows the quantity

350: $({F_{\rm rad}}/{F_{\rm grav}})\times ({M_{11}}/{\dot{N}_{\alpha,54}})$, where $M_{11}=(M/10^{11} M_{\odot})$. For example, if $M_{11}=0.1$ then

351: Figure~\ref{fig:force1} shows that radiation pressure exceeds gravity

352: out to $r=100$ kpc, well beyond the virial radius of a halo of this mass at $r_{\rm vir}\sim 6.6$ kpc.

353: 

354: Most importantly, Figure~\ref{fig:force1} shows that the two approaches

355: used to compute the radiation force in the simulation yield consistent

356: results, with a noticeable deviation only at the largest radii ($r\sim 1$

357: Mpc). At large radii most photons stream outwards radially and the

358: Eddington approximation that was used to derive Eq.~(\ref{eq:flux2})

359: becomes increasingly unreliable.

360: 

361: Next, we use the radiative transfer code to explore the magnitude of the

362: Ly$\alpha$ radiation pressure for a range of models which represent an

363: evolutionary sequence of structure formation in the Universe. We focus on

364: the Ly$\alpha$ radiation pressure on gas surrounding ({\it i}) the first

365: stars (\S~\ref{sec:firststar}); ({\it ii}) the first galaxies

366: (\S~\ref{sec:firstgalaxy}); and ({\it iii}) the interstellar medium of

367: galaxies (\S~\ref{sec:wind}).

368: 

369: \section{Results}

370: \label{sec:result}

371: 

372: \subsection{Case I: A Single Massive Star in a Minihalo}

373: \label{sec:firststar}

374: \begin{figure*}

375: \vbox{\centerline{\epsfig{file=fig3.eps,angle=270,width=15.0truecm}}}

376: \caption[]{The energy density in the Ly$\alpha$ radiation field ({\it

377: left panel}), and the ratio between the radiation and the

378: gravitational forces ({\it right panel}) for a single very massive

379: star ($M_*=100M_{\odot}$) in a minihalo $M_h=2\times 10^6M_{\odot}$

380: . The {\it red solid} line represents a model in which the IGM is

381: at the mean cosmic density and undergoes Hubble expansion right

382: outside the virial radius at $r=0.26$ kpc. The {\it blue dotted line}

383: shows a more realistic model in which the IGM is overdense near the

384: virial radius, and in which the intergalactic gas is gravitationally

385: pulled towards the minihalo (see text). The {\it black dashed line}

386: shows the same model as the {\it red solid line} but with the neutral

387: fraction increasing linearly between $r_{\rm vir}$ and $2r_{\rm

388: vir}$. The star ionizes all the gas out to the virial

389: radius. Ly$\alpha$ photons freely propagate until they reach the edge

390: of the HII region, where they are likely to be scattered back into the

391: ionized minihalo. This yields an almost constant radiation energy

392: density. Once outside the HII region, the radiation force dominates

393: over gravity out to $r=10$ kpc and may accelerate neutral gas outside

394: the HII region to velocities of order $\sim 10$ km s$^{-1}$.}

395: \label{fig:caseI}

396: \end{figure*}

397: Numerical simulations of structure formation suggest the first stars that

398: formed in our Universe were massive ($M_\star \sim 100M_{\odot}$), and

399: formed as single objects in dark matter halos with masses of $M\sim

400: 10^6M_{\odot}$ that collapsed at $z>10$

401: \citep[e.g.][]{Haiman96,Abel02,Yoshida06}. Here, we focus our attention on

402: a star with a mass $M_*=100M_{\odot}$ that formed at $z=15$ in a dark

403: matter mini-halo of mass $M=2\times 10^6M_{\odot}$. The star emits

404: $10^{50}$ ionizing photons per second \citep{S02,Abel07}. We assume that

405: the ionizing flux ionizes all the gas out to the virial radius of the dark

406: matter halo ($r_{\rm vir}=0.26$ kpc) but not beyond that radius

407: \citep{Kita04}. Hence, the IGM gas surrounding this central source is

408: assumed to be neutral ($x_{\rm HI}=1.0$) and cold ($T_{\rm gas}=300$K, which

409: corresponds to the temperature of the neutral IGM at $z=15$ due to X-Ray heating, see e.g. Fig~1 of Pritchard \& Loeb 2008).

410:  

411: Recombination following photoionization converts $\sim 68\%$ of all

412: ionizing photons into Ly$\alpha$ photons \citep[][p

413: 387]{Osterbrock89}. Hence, the entire halo is a Ly$\alpha$ source that is

414: surrounded by neutral intergalactic gas. To determine the radial dependence

415: of the Ly$\alpha$ production rate ($S_{\nu}$ in Eq.~\ref{eq:RT}), we need

416: to specify the gas density profile. We assume that the gas distribution

417: inside the dark matter halo is described by an NFW-profile with a

418: concentration parameter $C=5$ and a thermal core\footnote{With this

419: gas density profile, the total recombination rate inside the dark matter

420: halo is $\int_0^{r_{\rm vir}}dr\hs 4\pi r^2n_{\rm H}^2\alpha_{\rm rec}\sim

421: 4.5\times 10^{49}$ s$^{-1}$. The total recombination rate can be increased

422: to balance the photoionization rate by introducing a clumping factor

423: $K\equiv \langle n_H^2 \rangle/\langle n_{\rm H} \rangle^2\sim 2$.} at

424: $r<3r_{\rm vir}/4C$ \citep[see][]{Maller04}. We point out however, that our

425: final results are not sensitive to our choice of $S_{\nu}(r)$.

426: 

427: Once $S_{\nu}(r)$ has been determined, we find the radius, $r$, at which a

428: Ly$\alpha$ photon is generated in the Monte-Carlo simulation from the

429: relation

430: \begin{equation}

431: R=\frac{1}{N}\int_0^{r}dr\hs 4\pi r^2 n_{\rm H}^2\alpha_{\rm rec},

432: \label{eq:s}

433: \end{equation} 

434: where $R$ is a random number between $0$ and $1$, $N=\int_0^{r_{\rm

435: vir}}dr\hs 4\pi r^2 n_{\rm H}^2\alpha_{\rm rec}$ is the total recombination

436: rate inside the dark matter halo, and $\alpha_{\rm rec}=2.6\times 10^{-13}$

437: cm$^3$ s$^{-1}$ is the case-B recombination coefficient at a temperature

438: $T=10^4$ K \citep[e.g.][]{Hui97}. Once the photon is generated, it scatters

439: through the neutral IGM until it has redshifted far enough from resonance

440: that it can escape to the observer.

441: 

442: In the {\it left panel} of Figure~\ref{fig:caseI} we show the energy

443: density (in erg cm$^{-3}$) of the Ly$\alpha$ radiation field as a function

444: of radius. The {\it red solid} line represents a model in which we assumed

445: the IGM to follow the mean density and Hubble expansion right outside the

446: virial radius. The {\it blue dotted line} shows a more realistic model in

447: which the IGM is still overdense near the virial radius, and in which the

448: intergalactic gas is gravitationally pulled towards the minihalo (see

449: Dijkstra et al 2007 for a quantitative description of the density and

450: velocity profiles based on the model of Barkana 2004). The {\it black

451: dashed line} shows the same model as the {\it red solid line} but with the neutral fraction increasing

452: linearly between $r_{\rm vir}$ and $2r_{\rm vir}$. This provides a better

453: representation of the fact that the central population III star emits

454: ionizing photons with energies $\gsim 54$ eV, which can photoionize

455: hydrogen (and helium) atoms that lie deeper in the IGM. The goal of this

456: model is to investigate whether our results depend sensitively on the

457: presence of a sharp boundary between HI and HII.

458: 

459: All models show that the radiation energy density within the fully ionized

460: minihalo ($r\lsim r_{\rm vir}=0.26$ kpc) has only a weak dependence on

461: radius, i.e. $d\log U/d\log r \gsim -1$. Naively, this may appear

462: surprising given the fact that within the model, no scattering occurs

463: within the virial radius and one may expect the energy density in the

464: Ly$\alpha$ radiation field to scale as $U \propto r^{-2}$. However, in

465: reality $U$ obtains only a weak radial dependence because the radiation can be

466: scattered back into the ionized minihalo as soon as it 'hits' the wall of

467: neutral IGM gas. Ly$\alpha$ photons are therefore trapped inside the

468: ionized minihalo and their energy density is boosted to a value that is

469: only weakly dependent on radius. On the other hand, for $r\gsim r_{\rm

470: vir}$ we find that $d\log U/d\log r \lsim -2$, which is because Ly$\alpha$

471: photons are trapped more efficiently near the edge of the HII region, while

472: they stream freely outwards at larger radii (as in \S~\ref{sec:test} and

473: Fig~\ref{fig:uden}). Figure~\ref{fig:caseI} shows clearly that the radial

474: dependence of the Ly$\alpha$ energy density is not sensitive to the

475: detailed model assumptions about the gas in the IGM.

476: 

477: In the {\it right panel} of Figure~\ref{fig:caseI} we show the ratio

478: between the radiation force (Eq.~\ref{eq:force}) and the gravitational

479: force on a single hydrogen atom), $F_{\rm grav}={GM(<r)m_{\rm

480: p}}/{r^{2}}$, where $M(<r)$ is the total (baryons + dark matter) mass

481: enclosed within a radius $r$. In all models, radiation pressure dominates

482: over gravity by as much as $\gsim 2$ orders of magnitude. The radiation

483: force is largest for the models in which the IGM is assumed to be at

484: mean density, because of the $n_{\rm H}^{-1}$ factor in the

485: equation for the radiation force (Eq.~\ref{eq:force}). Note that the spike

486: near $r\sim 2.6$ kpc for the other two models is due to an artificial

487: discontinuity in the IGM velocity field that exists in this model. 

488: 

489: \begin{figure*}

490: \vbox{\centerline{\epsfig{file=fig4.eps,angle=270,width=15.0truecm}}}

491: \caption[]{Same as Figure~\ref{fig:caseI}, but for the case of a star

492: forming galaxy ($\dot{M}_*=0.34M_{\odot}$ yr$^{-1}$, see text), surrounded

493: by an HII region with a radius $R_{\rm HII}=50$ kpc ($R_{\rm HII}=20$ kpc)

494: for the {\it solid line} ({\it dotted line}), which is in turn surrounded

495: by a fully neutral intergalactic medium (IGM). For the assumed total halo

496: mass of $M_{\rm tot}=10^9M_{\odot}$, the pressure exerted by the Ly$\alpha$

497: photons is not large enough to exceed gravity. However, radiation pressure

498: wins for $M_{\rm tot}=10^8M_{\odot}$, but even in this case the radiation

499: force is not large enough to produce a significant wind speed in the IGM

500: (see text). }

501: \label{fig:caseII}

502: \end{figure*}

503: 

504: Ly$\alpha$ radiation pressure may operate throughout the lifetime of the

505: central star. Over a lifetime of $\sim 2.5$ Myr \citep[see Table 4

506: of][]{S02}, this mechanism is capable of accelerating the gas to velocities

507: of $10$ ($50$) km s$^{-1}$ at $r=0.3$ kpc in the model represented by the

508: {\it blue dotted} ({\it red solid}) {\it line}, and to $4$ ($16$) km

509: s$^{-1}$ at $r=0.4$ kpc (the reason for this large difference is that the edge of the HII region lies at $r=0.26$ kpc. Hence, gas at $r=0.3$ kpc is separated by 0.04 kpc from this edge, while gas at $r=0.4$ kpc is separated by a distance that is 3.5 times larger). 

510:  

511: Thus, Ly$\alpha$ radiation pressure can accelerate the gas to velocities

512: that exceed the escape velocity from the dark matter halo ($v_{\rm esc}\sim

513: \sqrt{2}v_{\rm circ}\sim 8$ km s$^{-1}$) as well as the sound speed of the

514: intergalactic medium ($c_{\rm s}=2.2(T_{\rm gas}/300\hs{\rm K})^{1/2}$ km

515: s$^{-1}$). 

516: 

517: Note the as the gas is pushed out and its velocity profile changes, the

518: subsequent radiative transfer is altered. For example, we repeated the

519: radiative transfer calculation for models in which gas at $r_{\rm vir}< r

520: \lsim 2r_{\rm vir}$ was accelerated to velocities in the range $10-20$ km

521: s$^{-1}$ (outward) and found a slightly shallower profile for $U(r)$ which lowered

522: the radiation force by a factor of $\sim 3$. Consequently, the acceleration

523: of the gas decreases with time and the actual velocities reached by the gas

524: are lower than the estimates given above by a factor of a few.

525: Nevertheless, the resulting velocities are still substantial.

526: 

527: Our calculations imply that Ly$\alpha$ radiation pressure can affect the

528: gas dynamics in the IGM surrounding minihalos that contain the first

529: stars. The impact of Ly$\alpha$ radiation pressure increases with

530: decreasing density of the surrounding gas in the IGM. In practice, the

531: distribution of the IGM is not spherically symmetric. Instead, the density

532: is expected to vary from sightline to sightline (being large along

533: filaments and small along voids). Our results imply that Ly$\alpha$

534: radiation pressure will be most efficient in 'blowing out' the lower

535: density gas. This conjecture is supported by the tendency of Ly$\alpha$

536: photons to preferentially scatter through the low-density gas; their

537: propagation along the path of least resistance would naturally boost up the

538: Ly$\alpha$ flux there. This effect will be moderated by the tendency of the

539: HII region around the first stars to extend further into the low density

540: gas (in 'butterfly'-like patterns, e.g. Abel et al, 1999).

541: 

542: If the central star dies in a supernova explosion, then the resulting

543: violent outflow could blow most of the baryons out from the

544: minihalo. However, stars with masses in the range $30M_{\odot}\lsim

545: M_{*}\lsim140M_{\odot}$ and $M_*\gsim 260M_{\odot}$, are not expected to

546: end their lives in a supernova. Instead, these stars collapse directly to a

547: black hole \citep{HW02} and have weak winds (because of the lack of

548: heavy elements in their atmosphere), so that radiation pressure may

549: be the dominant process that affects their surrounding IGM.

550: 

551: In summary, Ly$\alpha$ radiation pressure on the neutral IGM around

552: minihalos in which the first stars form, can exceed gravity by orders of

553: magnitude and launch supersonic winds. Our limited analysis does not allow

554: a detailed discussion on the consequences of these winds. This requires 3D

555: simulations with cosmological initial conditions that capture the full IGM

556: density field around the minihalo and that track the evolution of the

557: shocks that may form in the IGM. Such simulation are numerically

558: challenging as they require self-consistent treatment of gas dynamics and

559: Ly$\alpha$ radiative transfer in a moving inhomogeneous medium.

560: 

561: \subsection{Case II: A Young Star Forming Galaxy}

562: \label{sec:firstgalaxy}

563: 

564: \begin{figure*}

565: \vbox{\centerline{\epsfig{file=fig5.eps,angle=270,width=15.0truecm}}}

566: \caption[]{Same as in Figure~\ref{fig:caseI} and Figure~\ref{fig:caseII},

567: but for models in which a central Ly$\alpha$ source of luminosity

568: $L_{\alpha}$ is surrounded by a thin ($r_{\rm sh,min}=0.9r_{\rm sh,max}$)

569: spherical shell of HI that is expanding at $v_{\rm sh}=200$ km

570: s$^{-1}$. The {\it blue dotted lines} ({\it red solid lines}) represent a

571: model in which $N_{\rm HI}=10^{21}$ cm$^{-2}$ and $r_{\rm sh,max}=0.1$ kpc

572: ($N_{\rm HI}=10^{19}$ cm$^{-2}$ and $r_{\rm sh,max}=1.0$ kpc). When

573: calculating $F_{\rm grav}$, we assumed the total mass

574: enclosed by the supershell to be $10^8M_{\odot}$ (see text). The results

575: strongly suggest that Ly$\alpha$ radiation pressure may be dynamically

576: important in the interstellar medium of galaxies.  The {\it total}

577: radiation force on the shell (obtained as a sum over all atoms) may be

578: computed via $F_{\rm tot}=M_F L_{\alpha}$/c, where $M_F$ may be thought of

579: as a force multiplication factor that depends both on the shell's outflow

580: speed, $v_{\rm exp}$, and its HI column density, $N_{\rm HI}$. The

581: dependence of $M_F$ on these parameters is shown in Fig~\ref{fig:fscat}.}

582: \label{fig:caseIII}

583: \end{figure*}

584: 

585: Our second case concerns a young galaxy that is forming multiple stars in a

586: dark matter halo of mass $M=10^9 M_{\odot}$ at $z=10$. We assume that the

587: galaxy is converting a fraction $f_*=10\%$ of its baryons into stars over

588: $\sim 0.1t_{\rm H}$ \citep[][]{Wyithe06}, where $t_{\rm

589: H}=[{2}/{3H(z)}]\sim0.49$ Gyr, is the age of the Universe at $z=10$. This

590: translates to a star formation rate of $\dot{M}_*=0.34M_{\odot}$

591: yr$^{-1}$. For population III stars forming out of pristine gas, the total

592: emission rate of ionizing photons is $\dot{N}_{\rm ion}\sim 3\times

593: 10^{53}$ s$^{-1}$ \citep{S02}\footnote{More precisely, the ionizing photon

594: production rate is $\sim 10^{54}$ s$^{-1}$ $\times({\rm

595: SFR}/M_{\odot}\hs{\rm yr}^{-1})$ in the no-mass-loss model of Schaerer

596: (2002) in which metal-free stars form according to a Salpeter IMF with

597: $M_{\rm low}=1M_{\odot}$ and $M_{\rm high}=500 M_{\odot}$ (his model

598: 'B').}. If $\sim 1\%$ of the ionizing photons escape from the galaxy

599: \citep[][]{Chen07,Gnedin08}, then this translates to a Ly$\alpha$

600: luminosity of $L_{\alpha}=3\times 10^{42}$ erg s$^{-1}$. Furthermore, this

601: galaxy can photoionize a spherical HII region of a radius $R_{\rm HII}\sim

602: 50$ physical kpc. Note however, that other ionizing sources

603: would likely exist within this HII region. Indeed, clusters of sources are

604: thought to determine the growth of ionized bubbles during

605: reionization. This results in a characteristic HII region size that is

606: significantly larger than that produced by single source, especially during

607: the later stages of reionization \citep[e.g.][]{F04,McQuinn07}. In this

608: framework, our model represents a star forming galaxy during the early

609: stages of reionization or alternatively a galaxy that lies $50$ kpc away

610: from the edge of a larger ionized bubble.

611: 

612: In this particular case, the majority of all recombination events occur in

613: the central galaxy. Thus, we initiate all Ly$\alpha$ photons at $r=0$ in

614: the Monte-Carlo simulation. We assume that the gas is completely ionized

615: out to $R_{\rm HII}=50$ kpc, beyond which it is neutral. As 

616: shown in \S~\ref{sec:firststar}, this abrupt transition in the ionized

617: fraction of H in the gas does not affect our results.

618: 

619: The {\it left panel} of Figure~\ref{fig:caseII} shows the energy density

620: (in erg cm$^{-3}$) of the Ly$\alpha$ radiation field as a function of

621: radius. The {\it solid line} represents the model discussed above. A kink

622: in the energy density is seen at the edge of the HII region (see

623: \S~\ref{sec:firststar} for a more detailed discussion of the profile). The

624: {\it dotted line} represents a variant of the model in which we have

625: reduced the size of the HII region to $R_{\rm HII}=20$ kpc.

626:  

627: The {\it right panel} of Figure~\ref{fig:caseII} shows the ratio between

628: the radiation and the gravitational forces on a single hydrogen atom. In

629: our fiducial model, the radiation force does not exceed gravity; rather, at

630: the edge of the HII region, gravity is $\sim 3$ times stronger. The

631: radiation force becomes equal to the gravitational force if $R_{\rm

632: HII}=20$ kpc. This requires an extremely low [by a factor $\sim (50/20)^3$]

633: escape fraction of ionizing photons, $f_{\rm esc}\sim 6\times10^{-4}$.

634: 

635: Alternatively, radiation pressure {\it is} important when the halo mass of

636: the star forming region is reduced to $10^8M_{\odot}$. Halos of this mass are the the most abundant halos at $z\sim 10$ that are capable of cooling via excitation of atomic hydrogen (i.e. their virial

637: temperature just exceeds $T_{\rm vir}\sim 10^4$ K, e.g. Barkana \& Loeb

638: 2001). The total gas reservoir inside these halos is $M_{\rm

639: b}=\frac{\Omega_{\rm b}}{\Omega_{\rm m}}M_{\rm tot}\sim 1.5\times 10^7

640: M_{\odot}$, and so these halos can sustain a star formation rate of

641: $\dot{M}_{*}=0.3M_{\odot}$ yr$^{-1}$ for up to $\sim 50$ Myr. However, even

642: if the radiation force is allowed to operate for $\sim 50$ Myr, we find

643: that radiation pressure cannot accelerate the gas in the IGM to velocities

644: that exceed $\sim 1$ km s$^{-1}$. We therefore conclude that although

645: Ly$\alpha$ pressure may exceed gravity in the neutral IGM that surrounds

646: HII regions around $M_{\rm tot}=10^8M_{\odot}$ halos, the absolute

647: magnitude of the radiation force is too weak to drive the IGM to supersonic

648: velocities.

649: 

650: \subsection{Case III: Ly$\alpha$ Driven Galactic Supershells}

651: \label{sec:wind}

652: 

653: In principle, Ly$\alpha$ radiation pressure can be important when neutral

654: gas exists in close proximity to a luminous Ly$\alpha$ source. So far, we

655: focused our attention on HI gas in the IGM. However, neutral gas in the

656: interstellar medium (ISM) of the host galaxy is located closer to the

657: Ly$\alpha$ sources and should be exposed to an even stronger Ly$\alpha$

658: radiation pressure.  Indeed, it has been demonstrated

659: \citep[e.g.][]{Ahn02,Verhamme08} that scattering of Ly$\alpha$ photons by

660: neutral hydrogen atoms in a thin (with a thickness much smaller than its radius), outflowing 'supershell' of HI gas surrounding

661: the star forming regions can naturally explain two observed phenomena:

662: ({\it i}) the common shift of the Ly$\alpha$ emission line towards the red

663: relative to metal absorption lines and the host galaxy's systemic redshift determined from other nebular recombination lines \citep[e.g.][]{Pettini01,Shapley03}; and ({\it ii}) the asymmetry of the Ly$\alpha$ line with emission extending well into its red wing \citep[e.g.][]{L95,Tapken07}.

664: 

665: The existence of thin, outflowing shells of neutral atomic hydrogen around HII regions is confirmed by

666: HI-observations of our own Milky-Way \citep[][]{Heiles84} and other nearby

667: galaxies \citep[e.g.][]{Ryder95}. The largest of these shells, so-called

668: 'supershells', have radii of $r_{\rm max}\sim 1$ kpc

669: \citep[e.g][]{Ryder95,M02} and HI column densities in the range $N_{\rm

670: HI}\sim 10^{19}$--$10^{21}$ cm$^{-2}$

671: \citep[e.g.][]{L95,Kunth98,Verhamme08}. Supershells are thought to be

672: generated by stellar winds or supernovae explosions which sweep-up gas into

673: a thin expanding neutral shell \citep[see e.g.][for a review]{T88}. The

674: back-scattering mechanism attributes both the redshift and asymmetry of the

675: Ly$\alpha$ line to the Doppler boost that Ly$\alpha$ photons undergo as

676: they scatter off the outflow on the far side of the galaxy back towards the

677: observer \citep[e.g.][]{Lee98,Ahn02,Ahn03,Ahn04,Verhamme06,Verhamme08}. It is interesting to investigate whether Ly$\alpha$ radiation

678: pressure may provide an alternative mechanism that determines the

679: supershell kinematics.

680: 

681: In Figure~\ref{fig:caseIII} we show the energy density ({\it left panel})

682: and the Ly$\alpha$ radiation force ({\it right panel}) for two models. Both

683: models assume that: ({\it i}) there is a Ly$\alpha$ source at $r=0$ with a

684: luminosity of $L_{\alpha}=10^{43}$ erg s$^{-1}$; ({\it ii}) the emitted

685: Ly$\alpha$ spectrum prior to scattering has a Gaussian shape as a function

686: of photon frequency with a Doppler velocity width of $\sigma=50$ km

687: s$^{-1}$; ({\it iii}) the spatial width of the supershell is $10\%$ of its

688: radius; and ({\it iv}) the shell has an outflow velocity of $v_{\rm}=200$

689: km s$^{-1}$. The {\it blue dotted (red solid) lines} represent a model in

690: which the supershell has a column density of $N_{\rm HI}=10^{21}$

691: ($N_{\rm HI}=10^{19}$) cm$^{-2}$ and a maximum radius that is $r_{\rm

692: sh}=0.1$ kpc ($r_{\rm sh}=1.0$) kpc. Our calculations assume that there is no neutral gas (or dust) interior to the HI supershell.

693: \begin{figure}

694: \vbox{\centerline{\epsfig{file=fig6.eps,angle=270,width=9.0truecm}}}

695: \caption[]{The multiplication factor $M_F$ provides the total force that

696: Ly$\alpha$ photons exert on a spherical HI shell, $F_{\rm tot}=M_F

697: L_{\alpha}/c$, where $L_{\alpha}$ is the Ly$\alpha$ luminosity of the

698: source in erg s$^{-1}$ and $c$ is the speed of light. The plot shows $M_F$

699: as a function of the expansion velocity of the HI shell, $v_{\rm sh}$, for

700: three values of HI column density, and under the assumption that the HI shell surrounds an empty cavity. The parameter $M_F$ provides a measure

701: of the efficiency by which Ly$\alpha$ photons can be 'trapped' by the shell

702: of HI gas. Thus, $M_F$ increases with increasing $N_{\rm HI}$ and

703: decreasing $v_{\rm sh}$. The {\it dotted horizontal lines} show the values of

704: $M_F$ that have been derived in the past (for a static, uniform,

705: infinite slab of material e.g. Adams 1975).}

706: \label{fig:fscat}

707: \end{figure}

708: 

709: The {\it left panel} of Figure~\ref{fig:caseIII} shows that inside the

710: supershell the energy density decreases more gradually than $r^{-2}$

711: because of photon trapping (similarly to the previously discussed cases in

712: \S~\ref{sec:firststar}-\S~\ref{sec:firstgalaxy}). The shell with the larger

713: column of HI is more efficient at trapping the Ly$\alpha$ photons, and thus

714: yields a flatter energy density profile. In both models the energy density

715: drops steeply within the supershell (the energy density decreases as $r^{-2}$ outside the shell, if no scattering occurs here).

716: 

717: The {\it right panel} of Figure~\ref{fig:caseIII} shows the ratio between

718: the radiation and gravitational forces.  Towards the center of the dark

719: matter halo, baryons dominate the mass density and an evaluation of $F_{\rm

720: grav}$ requires assumptions about the radial distribution of the

721: baryons. For simplicity, we consider a fixed total mass interior to the

722: supershell of $M(<r)=10^8M_{\odot}$, so that $F_{\rm

723: grav}(r)=GM(<r)m_{p}/r^{2}$. Note that any assumed mass profile $M(<r)$

724: will not affect the results as long as $F_{\rm rad}\gg F_{\rm grav}$.

725: 

726: We find that the radiation force exceeds gravity in both examples under

727: consideration. For $N_{\rm HI}=10^{21}$ cm$^{-2}$ and $r_{\rm sh}=0.1$ kpc,

728: $F_{\rm rad}\sim 10 F_{\rm grav}$. Thus, radiation pressure would have been

729: important even if we had chosen $M\sim 10^{9}M_{\odot}$. 

730: %The virial radius of a dark matter halo of mass $M_{\rm tot}$ that virialized at redshift $z$ is given by $r_{\rm vir}=2$ kpc$(M_{\rm tot}/5\times 10^{7}\hs M_{\odot})^{1/3}([1+z]/6)^{-1}$ \citep[see, e.g. Eq. 24 in][]{BL01}. Even if we assume that a fraction $f_b=\frac{\Omega_{\rm b}}{\Omega_{\rm m}}\sim 0.15$ of the total mass, corresponding to a mass of $\sim 10^7 M_{\odot}$, resides in a sphere of radius $r_{\rm b}=\lambda r_{\rm vir}=0.1$ kpc (where $\lambda\sim 0.05$ is the spin-parameter), it is still not possible for . 

731: In the model with $N_{\rm HI}=10^{19}$ cm$^{-2}$ and $r_{\rm sh}=1.0$ kpc,

732: $F_{\rm rad}\sim 10^2 F_{\rm grav}$, and radiation pressure would have been

733: important even if $M\sim 10^{10}M_{\odot}$. Hence, our calculations

734: strongly suggest that Ly$\alpha$ radiation pressure may be dynamically

735: important in the ISM of galaxies.

736: 

737: The {\it total} Ly$\alpha$ radiation force is obtained by summing the force

738: over all atoms in the supershell. It is interesting to compare this force

739: to $L_{\alpha}/c$. The latter quantity denotes the total momentum transfer

740: rate (force) from the Ly$\alpha$ radiation field to the supershell under

741: the assumption that each Ly$\alpha$ photon is re-emitted isotropically

742: after entering the shell (including multiple scatterings inside the

743: supershell). In Figure~\ref{fig:fscat} we plot the quantity $M_F$ which is

744: defined as

745: \begin{equation}

746: M_F\equiv \frac{\sum_{\rm atoms}F_{\rm rad}}{L_{\alpha}/c},

747: \end{equation} 

748: as a function of the expansion velocity of the shell, $v_{\rm sh}$ for

749: three different values of $N_{\rm HI}$.

750: 

751: Figure~\ref{fig:fscat} shows that $M_F$, which can be thought of as a

752: force multiplication factor\footnote{This term derives from the

753: (time-dependent) force-multiplication function $M(t)$ that was

754: introduced by \citet{Castor75}, as $F_{\rm rad}\equiv

755: M(t)({\tau_eL_{\rm bol}}/{c})$. Here, $F_{\rm rad}$ is the total

756: force that radiation exerts on a medium, $\tau_e$ is the total optical

757: depth to electron scattering through this medium. The function $M(t)$

758: arises because of the contribution of numerous metal absorption lines

759: to the medium's opacity, and can be as large as $M_{\rm max}(t)\sim

760: 10^3$ in the atmospheres of O-stars \citep{Castor75}.}, greatly

761: exceeds unity for low shell velocities and large HI column

762: densities. The parameter $M_F$ is related to the mean number of times

763: that a Ly$\alpha$ photon 'bounces' back and forth between opposite

764: sides of the expanding shell. For example, $M_F=1$ when all Ly$\alpha$

765: photons enter the shell, scatter once, and then escape from the shell

766: in no preferred direction. On the other hand, $M_F=3$ when all

767: Ly$\alpha$ photons enter the shell, scatter back towards the opposite

768: direction, and then escape in no preferred direction after scattering

769: in the shell for a second time. A schematic illustration of this

770: argument is provided in Figure~\ref{fig:scheme}. Note that when

771: $M_F=1$ ($M_F=3$), each photon spends on average a timescale of

772: $r_{\rm sh}/c$ ($3r_{\rm sh}/c$) in the bubble enclosed by the

773: shell. In other words, the factor $M_F$ relates to the 'trapping

774: time', $t_{\rm trap}$, that denotes the total time over which

775: Ly$\alpha$ photons are trapped inside the supershell\footnote{This

776: argument ignores the time spent on scattering inside the supershell

777: itself. Photons penetrate on average an optical depth $\tau=1$ into

778: the shell. If this corresponds to a physical distance that is

779: significantly smaller than the thickness of the shell (denoted by

780: $\Delta r_{\rm sh}$), then only a tiny fraction of the photons will

781: diffuse through the shell. Hence, when averaged over these photons,

782: ignoring the time spent inside the supershell itself is

783: justified. Alternatively, photons with a mean free path that is at

784: least comparable to the thickness of the shell, only spend a time

785: $\sim \Delta r_{\rm sh}/c \ll r_{\rm sh}/c$ inside the supershell,

786: which provides a negligible contribution to the trapping time.} (see

787: \S~\ref{sec:intro}) through the relation $M_F=t_{\rm trap}/(r_{\rm

788: sh}/c)$. Indeed, when $v_{\rm sh}\rightarrow 0$ we find that $M_F$

789: reproduces the value $15(\tau_0/10^{5.5})$ (indicated by {\it

790: horizontal dotted lines}) that was found by \citet{Adams75} and

791: \citet{Bo79} reasonably well (keeping in mind that these authors

792: derived their result for a static, uniform, infinite slab of material,

793: and assumed different frequency distributions for the emitted

794: Ly$\alpha$ photons).

795: 

796: \begin{figure}

797: \vbox{\centerline{\epsfig{file=fig7.eps,angle=0,width=6.0truecm}}}

798: \caption[]{A schematic illustration of the origin of the force

799: multiplication factor $M_F$ in an expanding supershell. The Ly$\alpha$

800: source is located at the center of the expanding HI supershell. The {\it

801: solid line} represents the trajectories of photons that enter the shell,

802: scatter once, and then escape from the shell isotropically (indicated by

803: the {\it dashed lines}). This corresponds to $M_F=1$. On the other hand,

804: the {\it dot-dashed line} represents the trajectories of photons that enter

805: the shell, are scattered back in the opposite direction, and escape with no

806: preferred direction after scattering in the shell for a second time. This

807: corresponds to a case with $M_F=3$. In general, $M_F=t_{\rm trap}/(r_{\rm

808: sh}/c)$.}

809: \label{fig:scheme}

810: \end{figure}

811: \section{Conclusions}

812: \label{sec:conc}

813: 

814: We have applied an existing Monte-Carlo Ly$\alpha$ radiative transfer code

815: \citep[described and tested extensively in][]{mc} to the calculation of the

816: pressure that is exerted by Ly$\alpha$ photons on an optically thick

817: medium. This code enabled us to perform (the first) direct, accurate calculations of Ly$\alpha$ radiation pressure, which distinguishes this work from previous

818: discussions on the importance of Ly$\alpha$ radiation pressure in various

819: astrophysical environments.

820: 

821: We have focused on a range of models which represent galaxies at different

822: cosmological epochs. In \S~\ref{sec:firststar} we have shown that the

823: Ly$\alpha$ radiation pressure exerted on the neutral intergalactic medium

824: (IGM) surrounding minihalos ($M_{\rm tot}\sim 10^6 M_{\odot}$) in which the

825: first stars form, can exceed gravity by 2--3 orders of magnitude

826: (Fig~\ref{fig:caseI}), and in principle accelerate the gas in the IGM to

827: tens of km s$^{-1}$. Thus, Ly$\alpha$ radiation pressure can launch

828: supersonic winds in the IGM surrounding the first stars. Our analysis did

829: not allow a detailed study of the consequences of these winds. A

830: comprehensive study would require numerical simulations that capture the

831: full IGM density field around minihalos in 3D and track the evolution of

832: the shocks that may form in the IGM together with the Ly$\alpha$ radiative

833: transfer. In this paper, we have also shown that Ly$\alpha$ radiation

834: pressure is important in the neutral IGM that surrounds the HII regions

835: produced by galaxies with a total halo mass of $M_{\rm tot}=10^8M_{\odot}$

836: (Fig~\ref{fig:caseII}. These are the lowest mass, and hence the most

837: abundant, halos in which gas can cool via atomic line excitation. Here,

838: however, the absolute magnitude of the radiation force is too weak to drive

839: the gas to supersonic velocities.

840:  

841: Finally, we have shown in \S~\ref{sec:wind} that the Ly$\alpha$ radiation

842: pressure exerted on neutral gas in the interstellar medium (ISM) of a

843: galaxy can also have strong dynamical consequences. In particular, we have

844: found that the Ly$\alpha$ radiation force exerted on an expanding HI

845: supershell can exceed gravity by orders of magnitude

846: (Fig~\ref{fig:caseIII}), for reasonable assumptions about the

847: gravitational force. It is therefore possible that Ly$\alpha$ radiation

848: pressure plays an important role in determining the kinematics of HI

849: supershells around starburst galaxies. We have demonstrated that the total

850: Ly$\alpha$ radiation force on a spherical HI supershell can be written as

851: $F_{\rm rad}=M_F L_{\alpha}/c$, where the 'force-multiplication factor'

852: $M_F$ relates to the average trapping time of Ly$\alpha$ photons in the

853: neutral medium. The factor $M_F$ can greatly exceed unity, as illustrated

854: by Fig~\ref{fig:fscat}. For comparison, the maximum possible radiation

855: force due to continuum radiation\footnote{

856: %Trapping of continuum photons by free electrons results in a similar force multiplication factor for electron column densities well in excess of $N_e\sim 10^{24}$ cm$^{-2}$. For comparison, HI column densities in supershells are $N_{\rm HI}\lsim 10^{21}$ cm$^{-2}$, and no observational evidence exists that supershells predominantly consists of ionized gas. 

857: Continuum radiation may exert a force on steady-state outflows

858: (with a constant mass ejection rate, $\dot{M}$) around late-type stars

859: that may significantly exceed $\dot{M}\Delta v\gg L_{\rm bol}/c$

860: \citep[][]{salpeter74,Ivezic95}. This is not because of 'trapping' of

861: continuum photons, but related to the propagation speed of the photons

862: and the wind. We similarly expect the radiative force of trapped

863: Ly$\alpha$ photons in steady-state outflows to potentially exceed

864: $\dot{M}\Delta v\gg M_FL_{\alpha}/c$ (and as argued in this paper, it

865: is possible that $M_FL_{\alpha}/c>L_{\rm bol}/c$).

866: %This is mainly because the total mechanical momentum change of a shell of material as it traverses a distance $dr$ (in a time $dt=dr/v$) must be compared to the total momentum in the radiation field that was emitted by the star during this time $dt=dr/v$. In some cases this may result in a radiative force as large as $\dot{M}\Delta v\sim 20L_{\rm bol/c}$\citep{ivezic95}.

867:  Note though that these steady-state outflows are clearly different

868: from those discussed in \S~\ref{sec:wind}, in which a well defined

869: thin shell HI gas is physically separated from the central Ly$\alpha$

870: source.}  is $L_{\rm bol}/c$, in which $L_{\rm bol}$ is the

871: bolometric luminosity of the central galaxy.  For a typical star

872: forming galaxy, $L_{\alpha}\sim 0.07L_{\rm bol}$ \citep[e.g.][]{PP67},

873: whereas for a galaxy that contains population III stars,

874: $L_{\alpha}\sim 0.24L_{\rm bol}$ \citep{S03}. Hence, the Ly$\alpha$

875: radiation pressure can dominate the maximum possible continuum

876: radiation pressure if $M_F\gsim 14$ (for a normal stellar population),

877: a threshold which is easily exceeded at large column densities of

878: relatively slow-moving HI shells (see Fig~\ref{fig:fscat}).

879: 

880: The possibility that Ly$\alpha$ radiation alone can result in a

881: radiation force that exceeds $L_{\rm bol}/c$ is

882: important. \citet{Murray05} have shown that the total momentum carried

883: by radiation from a star-forming region can exceed the total momentum

884: deposited by supernova explosions in it, and so galactic outflows

885: may be driven predominantly by continuum radiation pressure. We have

886: argued that Ly$\alpha$ radiation pressure may in some cases be even

887: more important than continuum radiation pressure, and thus provide the

888: dominant source of pressure on neutral hydrogen in the ISM.

889: 

890: The important implication of our last result is that Ly$\alpha$ radiation

891: pressure may drive outflows of HI gas in the ISM. Observations of local

892: starburst galaxies have shown that the presence of outflowing HI gas may be

893: required to avoid complete destruction of the Ly$\alpha$ radiation by dust

894: and to allow its escape from the host galaxies

895: \citep{Kunth98,Hayes08,Ostlin08,Atek08}. At high redshifts, the Ly$\alpha$

896: emission line of galaxies is often redshifted relative to other nebular

897: recombination lines (such as H$\alpha$) and metal absorption lines

898: \citep[e.g.][]{Pettini01,Shapley03}. Furthermore, the spectral shape of the

899: Ly$\alpha$ emission line is typically asymmetric, with emission extending

900: well into the red wing of the line \citep[e.g.][]{L95}. Both of these

901: observations can be explained simultaneously if the observed Ly$\alpha$

902: photons scatter off neutral hydrogen atoms in an outflowing 'supershell' of

903: HI gas that surrounds the star forming regions

904: \citep{L95,T99,Ahn03,Ahn04,Verhamme06,Verhamme08}. The possibility that

905: Ly$\alpha$ radiation pressure may be important in determining the

906: properties of expanding supershells is exciting, and is discussed in more

907: detail in a companion paper (Dijkstra \& Loeb 2008).

908: 

909: 

910: {\bf Acknowledgments} 

911: This work is supported by in part by NASA grant

912: NNX08AL43G, by FQXi, and by Harvard University funds. We thank Christian Tapken and an anonymous referee for helpful constructive comments.

913: 

914: 

915: \begin{thebibliography}{xx}

916: \expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi

917: \bibitem[Abel et al.(1999)]{Abel99} Abel, T., Norman, M.~L., 

918: \& Madau, P.\ 1999, \apj, 523, 66 

919: 

920: \bibitem[Abel et al.(2002)]{Abel02} Abel, T., Bryan, G.~L., 

921: \& Norman, M.~L.\ 2002, Science, 295, 93 

922: 

923: \bibitem[Abel et al.(2007)]{Abel07} Abel, T., Wise, J.~H., 

924: \& Bryan, G.~L.\ 2007, \apjl, 659, L87 

925: 

926: \bibitem[Adams(1975)]{Adams75} Adams, T.~F.\ 1975, \apj, 201, 

927: 350 

928: 

929: \bibitem[Ahn \& Lee(2002)]{Ahn02} Ahn, S.-H., \& Lee, H.-W.\ 2002, Journal of Korean Astronomical 

930: 

931: \bibitem[Ahn et al.(2003)]{Ahn03} Ahn, S.-H., Lee, H.-W., \& 

932: Lee, H.~M.\ 2003, \mnras, 340, 863 

933: 

934: \bibitem[Ahn(2004)]{Ahn04} Ahn, S.-H.\ 2004, \apjl, 601, L25 

935: 

936: \bibitem[Atek et al.(2008)]{Atek08} Atek, H., Kunth, D., 

937: Hayes, M., Ostlin, G., Mas-Hesse, J.~M., 

938: \& .\ 2008, ArXiv e-prints, 805, arXiv:0805.3501 

939: 

940: \bibitem[Barkana 

941: \& Loeb(2001)]{BL01} Barkana, R., \& Loeb, A.\ 2001, \physrep, 349, 125 

942: \bibitem[Barkana(2004)]{Barkana04} Barkana, R.\ 2004, \mnras, 

943: 347, 59 

944: 

945: \bibitem[Bithell(1990)]{Bithell90} Bithell, M.\ 1990, \mnras, 

946: 244, 738 

947: 

948: \bibitem[Bonilha et al.(1979)]{Bo79} Bonilha, J.~R.~M., Ferch, R., Salpeter, E.~E., Slater, G., \& Noerdlinger, P.~D.\ 1979, \apj, 233, 649 

949: 

950: \bibitem[Castor et al.(1975)]{Castor75} Castor, J.~I., Abbott, D.~C., \& Klein, R.~I., 1975, \apj, 195, 157 

951: 

952: \bibitem[Chandrasekhar(1945)]{Chandra45} Chandrasekhar, S.\ 1945, 

953: \apj, 102, 402 

954: %\bibitem[Chiu \& Draine(1998)]{1998astro.ph..3209C} Chiu, W.~A., \& Draine, B.~T.\ 1998, ArXiv Astrophysics e-prints, arXiv:astro-ph/9803209 

955: 

956: \bibitem[Chen et al.(2007)]{Chen07} Chen, H.-W., Prochaska, 

957: J.~X., \& Gnedin, N.~Y.\ 2007, \apjl, 667, L125 

958: 

959: \bibitem[Cox(1985)]{Cox85} Cox, D.~P.\ 1985, \apj, 288, 465

960:  

961: \bibitem[Dennison et al.(2005)]{D05} Dennison, B., Turner, 

962: B.~E., \& Minter, A.~H.\ 2005, \apj, 633, 309 

963: 

964: \bibitem[Dijkstra et al.(2006)]{mc} Dijkstra, M., Haiman, 

965: Z., \& Spaans, M.\ 2006, \apj, 649, 14 

966: 

967: \bibitem[Dijkstra et al.(2007)]{igm} Dijkstra, M., Lidz, 

968: A., \& Wyithe, J.~S.~B.\ 2007, \mnras, 377, 1175 

969: 

970: \bibitem[Dijkstra \& Loeb(2008)]{dl} Dijkstra, M., \& Loeb, A.

971: 2008, submitted to \mnras

972: 

973: \bibitem[Dijkstra et al.(2008)]{FS} Dijkstra, M., et al.

974: 2008, accepted to \mnras

975: 

976: \bibitem[Elitzur \& Ferland(1986)]{EF86} Elitzur, M., \& Ferland, G.~J.\ 1986, \apj, 305, 35 

977: 

978: \bibitem[Furlanetto et al.(2004)]{F04} Furlanetto, S.~R., 

979: Zaldarriaga, M., \& Hernquist, L.\ 2004, \apj, 613, 1 

980: 

981: \bibitem[Gnedin et al.(2008)]{Gnedin08} Gnedin, N.~Y., Kravtsov, A.~V., \& Chen, H.-W.\ 2008, \apj, 672, 765 

982: 

983: \bibitem[Haehnelt(1995)]{Haenelt95} Haehnelt, M.~G.\ 1995, 

984: \mnras, 273, 249 

985: 

986: \bibitem[Haiman et al.(1996)]{Haiman96} Haiman, Z., Thoul, 

987: A.~A., \& Loeb, A.\ 1996, \apj, 464, 523 

988: 

989: \bibitem[Hayes et al.(2008)]{Hayes08} Hayes, M., Ostlin, G., 

990: Mas-Hesse, J.~M., \& Kunth, D.\ 2008, ArXiv e-prints, 803, arXiv:0803.1176 

991: 

992: \bibitem[Heger \& Woosley(2002)]{HW02} Heger, A., \& Woosley, S.~E.\ 2002, \apj, 567, 532 

993: 

994: \bibitem[Heiles(1984)]{Heiles84} Heiles, C.\ 1984, \apjs, 55, 

995: 585 

996: 

997: \bibitem[Hirata(2006)]{Hirata06} Hirata, C.~M.\ 2006, \mnras, 

998: 367, 259 

999: 

1000: \bibitem[Hui \& Gnedin(1997)]{Hui97} Hui, L., \& Gnedin, N.~Y.\ 1997, \mnras, 292, 27 

1001: 

1002: \bibitem[Ivezic \& Elitzur(1995)]{Ivezic95} Ivezic, Z., \& Elitzur, M.\ 1995, \apj, 445, 415

1003: 

1004: \bibitem[Kitayama et al.(2004)]{Kita04} Kitayama, T., Yoshida, 

1005: N., Susa, H., \& Umemura, M.\ 2004, \apj, 613, 631 

1006: 

1007: \bibitem[Komatsu et al.(2008)]{Komatsu08} Komatsu, E., et al.\ 

1008: 2008, ArXiv e-prints, 803, arXiv:0803.0547 

1009: 

1010: \bibitem[Kunth et al.(1998)]{Kunth98} Kunth, D., Mas-Hesse, J.~M.,

1011: Terlevich, E., Terlevich, R., Lequeux, J., \& Fall, S.~M.\ 1998, \aap, 334,

1012: 11

1013: 

1014: \bibitem[Lee 

1015: \& Ahn(1998)]{Lee98} Lee, H.-W., \& Ahn, S.-H.\ 1998, \apjl, 504, L61 

1016: 

1017: \bibitem[Lequeux et al.(1995)]{L95} Lequeux, J., Kunth, D., Mas-Hesse,

1018: J.~M., \& Sargent, W.~L.~W.\ 1995, \aap, 301, 18

1019: 

1020: \bibitem[Loeb \& Rybicki(1999)]{LR99} Loeb, A., \& Rybicki, G.~B.\ 1999,

1021: \apj, 524, 527 (LR99)

1022: 

1023: \bibitem[Loeb(2001)]{Loeb01} Loeb, A.\ 2001, \apjl, 555, L1 

1024: 

1025: \bibitem[Lucy(1999)]{Lucy99} Lucy, L.~B.\ 1999, \aap, 344, 282 

1026: 

1027: \bibitem[Lucy(2007)]{Lucy07} Lucy, L.~B.\ 2007, \aap, 468, 649 

1028: 

1029: \bibitem[Maller \& Bullock(2004)]{Maller04} Maller, A.~H., \& Bullock,

1030: J.~S.\ 2004, \mnras, 355, 694

1031: 

1032: \bibitem[McKee \& Tan(2008)]{McKee07} McKee, C.~F., \& Tan, J.~C.\ 2008, \apj, 681, 771 

1033: 

1034: \bibitem[McClure-Griffiths et al.(2002)]{M02} 

1035: McClure-Griffiths, N.~M., Dickey, J.~M., Gaensler, B.~M., 

1036: \& Green, A.~J.\ 2002, \apj, 578, 176 

1037: 

1038: \bibitem[McQuinn et al.(2007)]{McQuinn07} McQuinn, M., Lidz, A., Zahn, O., Dutta, S., Hernquist, L., 

1039: \& Zaldarriaga, M.\ 2007, \mnras, 377, 1043 

1040: 

1041: \bibitem[Murray et al.(2005)]{Murray05} Murray, N., Quataert, 

1042: E., \& Thompson, T.~A.\ 2005, \apj, 618, 569 

1043: 

1044: \bibitem[Oh \& Haiman(2002)]{Oh02} Oh, S.~P., \& Haiman, Z.\ 2002, \apj, 569, 558

1045: 

1046: \bibitem[Osterbrock(1989)]{Osterbrock89} Osterbrock, D.~E.\ 1989, {\it Astrophysics of gaseous nebulae and active galactic nuclei}, University of Minnesota, et al.~Mill Valley, CA, University Science Books.  

1047: 

1048: \bibitem[Ostlin et al.(2008)]{Ostlin08} Ostlin, G., Hayes, M., 

1049: Kunth, D., Mas-Hesse, J.~M., Leitherer, C., Petrosian, A., 

1050: \& Atek, H.\ 2008, ArXiv e-prints, 803, arXiv:0803.1174 

1051: 

1052: \bibitem[Ouchi et al.(2008)]{Ouchi08} Ouchi, M., et al.\ 2008, 

1053: \apjs, 176, 301 

1054: 

1055: \bibitem[Partridge \& Peebles(1967)]{PP67} Partridge, R.~B., \& Peebles, P.~J.~E.\ 1967, \apj, 147, 868 

1056: 

1057: \bibitem[Pettini et al.(2001)]{Pettini01} Pettini, M., Shapley, 

1058: A.~E., Steidel, C.~C., Cuby, J.-G., Dickinson, M., Moorwood, A.~F.~M., 

1059: Adelberger, K.~L., \& Giavalisco, M.\ 2001, \apj, 554, 981 

1060: 

1061: \bibitem[Pritchard \& Furlanetto(2006)]{PF06} Pritchard, J.~R., \& Furlanetto, S.~R.\ 2006, \mnras, 367, 1057 

1062: 

1063: \bibitem[Pritchard 

1064: \& Loeb(2008)]{PL08} Pritchard, J.~R., \& Loeb, A.\ 2008, ArXiv e-prints, 802, arXiv:0802.2102 

1065: 

1066: \bibitem[Rybicki \& Lightman(1979)]{RL79} Rybicki, G.~B., \& Lightman, A.~P.\ 1979, New York, Wiley-Interscience, 1979.~393 p.,  

1067: 

1068: \bibitem[Ryder et al.(1995)]{Ryder95} Ryder, S.~D., 

1069: Staveley-Smith, L., Malin, D., \& Walsh, W.\ 1995, \aj, 109, 1592 

1070: 

1071: \bibitem[Salpeter(1974)]{salpeter74} Salpeter, E.~E.\ 1974, \apj, 193, 585

1072: 

1073: \bibitem[Schaerer(2002)]{S02} Schaerer, D.\ 2002, \aap, 382, 28 

1074: 

1075: \bibitem[Schaerer(2003)]{S03} Schaerer, D.\ 2003, \aap, 397, 527 

1076: 

1077: \bibitem[Sethi et al.(2007)]{Sethi07} Sethi, S.~K., 

1078: Subrahmanyan, R., \& Roshi, D.~A.\ 2007, \apj, 664, 1 

1079: 

1080: \bibitem[Shapley et al.(2003)]{Shapley03} Shapley, A.~E., 

1081: Steidel, C.~C., Pettini, M., \& Adelberger, K.~L.\ 2003, \apj, 588, 65

1082:  

1083: \bibitem[Shimasaku et al.(2006)]{Shima06} Shimasaku, K., et 

1084: al.\ 2006, \pasj, 58, 313 

1085: 

1086: %\bibitem[Schneider et al.(2002)]{S02} Schneider, R., 

1087: %Ferrara, A., Natarajan, P., \& Omukai, K.\ 2002, \apj, 571, 30 

1088: 

1089: \bibitem[Tapken et al.(2007)]{Tapken07} Tapken, C., Appenzeller, I., Noll,

1090: S., Richling, S., Heidt, J., Meink{\"o}hn, E., \& Mehlert, D.\ 2007, \aap,

1091: 467, 63

1092: 

1093: \bibitem[Tenorio-Tagle \& Bodenheimer(1988)]{T88} Tenorio-Tagle, G., \&

1094: Bodenheimer, P.\ 1988, \araa, 26, 145

1095: 

1096: \bibitem[Tenorio-Tagle et al.(1999)]{T99} Tenorio-Tagle, 

1097: G., Silich, S.~A., Kunth, D., Terlevich, E., 

1098: \& Terlevich, R.\ 1999, \mnras, 309, 332 

1099: 

1100: \bibitem[Verhamme et al.(2006)]{Verhamme06} Verhamme, A., 

1101: Schaerer, D., \& Maselli, A.\ 2006, \aap, 460, 397 

1102: 

1103: \bibitem[Verhamme et al.(2008)]{Verhamme08} Verhamme, A., Schaerer, D.,

1104: Atek, H., \& Tapken, C.\ 2008, ArXiv e-prints, 805, arXiv:0805.3601

1105: 

1106: \bibitem[Wyithe \& Loeb(2006)]{Wyithe06} Wyithe, J.~S.~B., \& Loeb, A.\ 2006, \nat, 441, 322 

1107: 

1108: \bibitem[Yoshida et al.(2006)]{Yoshida06} Yoshida, N., Omukai, 

1109: K., Hernquist, L., \& Abel, T.\ 2006, \apj, 652, 6 

1110: 

1111: \end{thebibliography}

1112: 

1113: \appendix

1114: 

1115: \section{HI Level Populations}

1116: \label{app:n2}

1117: \subsection{The (De)Excitation Rates of the $2p$ Level}

1118: 

1119: Our calculations assumed that all neutral hydrogen atoms populate

1120: their electronic ground state (i.e. the $1s$ level). Below, we explore

1121: the physical conditions under which this assumption holds.

1122: 

1123: Processes that populate the $2p$ level include: ({\it i}) collisional

1124: excitation of neutral hydrogen atoms in both the $1s$ and $2s$ states

1125: by electrons and protons, ({\it ii}) recombination into the

1126: $2p$ state following photoionization or collisional ionization, ({\it

1127: iii}) photoexcitation of $2p$ level, ({\it iv}) photoexcitation of

1128: $np$ ($n>3$) levels, followed by a radiative cascade that passes

1129: through the $2p$ state. Processes that {\it de-populate} the $2p$ level

1130: include ({\it v}) collisional de-excitation to by electrons, ({\it vi})

1131: stimulated Ly$\alpha$ emission $2p\rightarrow 1s$ following the

1132: absorption of a Ly$\alpha$ photon, and ({\it vii}) spontaneous

1133: emission of a Ly$\alpha$ photon.

1134: 

1135: Quantitatively, the rate at which the population in the $2p$ level is

1136: populated is given by

1137: 

1138: \begin{eqnarray}

1139: \frac{dn_{2p}}{dt}=C_{1s2p}n_en_{1s}+ C_{\rm 2s2p}n_pn_{\rm 2s}+0.68n_en_{{\rm HII}}\alpha_{\rm rec,B}+\\ \nonumber

1140: \sum_{\rm n=2}^{\infty}P_{n}n_{1s}f_{\rm n2}-C_{2p1s}n_en_{2p}-C_{\rm 2p2s}n_pn_{2p}-P_{2}n_{2p}-n_{2p}A_{\rm 21},

1141: \label{eq:rate}

1142: \end{eqnarray} 

1143: where $C_{lu}$'s denote collisional excitation rate coefficients from

1144: level $l$ to $u$ (in cm$^3$ s$^{-1}$), $n_x$ denote number densities

1145: is species 'x' (i.e $n_e$ denotes the electron number density, while

1146: $n_{\rm HII}$ denotes the number density of HII ions), $P_{n}$ denote

1147: photoexcitation rates to the state $np$ (in s$^{-1}$), and $f_{n2}$

1148: denotes the probability that photoexcitation of the $np$ level results

1149: in a radiative cascade that passes through the $2p$ level.

1150: 

1151: In the reminder of this Appendix, we will estimate the

1152: order-of-magnitude of each term.

1153: 

1154: \begin{itemize}

1155: 

1156: \item The Einstein-A coefficient of the Ly$\alpha$ transition is $A_{21}=6.25\times 10^8$ s$^{-1}$. That is, spontaneous emission of Ly$\alpha$ depopulates the $2p$ state at a rate $A_{21}=6.25\times 10^8$ s$^{-1}$.

1157: 

1158: \item Collisional excitation of neutral hydrogen atoms in $1s$ level

1159: by electrons populates the $2p$ level at a rate

1160: $C_{1s2p}n_e=n_e\left[\frac{8.629\times10^{-6}}{T^{1/2}}\right]

1161: \left[\frac{\Omega(1s,2p)}{g_2}\frac{g_{2p}}{g_{1s}}\right]e^{-\chi/kT}~{\rm

1162: s^{-1}}$ \citep[e.g.][]{Osterbrock89}. Here, $T$ denotes the gas

1163: temperature in K, $n_e$ is the electron density in cm$^{-3}$, $g_{1s}=1$

1164: and $g_{2p}=3$ are the statistical weights of the $1s$ and $2p$ levels,

1165: $\Omega(1s,2p)=0.40-0.50$ \citep[5000 K $<$ T $<$ 2$\times 10^4$ K,][]{Osterbrock89}, and $\chi=10.2$ eV is

1166: the energy difference between the $1s$ and $2p$ levels. For

1167: temperatures $T< 2\times 10^4$ K, we find that $C_{1s2p}n_e<10^{-10}n_e$

1168: s$^{-1}$.

1169: 

1170: \item The collisional excitation rate of neutral hydrogen atoms in

1171: $2s$ level by protons (which dominate over collisions with electrons

1172: by about a factor of $\sim 10$) populates the $2p$ level at a rate

1173: $C_{2s2p}n_p\sim 2\times 10^{-3} n_p$ s$^{-1}$

1174: \citep[e.g.][]{Osterbrock89}. To assess the term $C_{\rm

1175: 2s2p}n_pn_{\rm 2s}$ requires one to compute $n_{2s}$. The fraction of

1176: atoms in the $2s$ state is determined by rates similar to those

1177: mentioned above, except that the $2s$-state is metastable and its

1178: Einstein coefficient is $A_{2s1s}\approx 8$ s$^{-1}$. The $2s$-state may

1179: therefore be overpopulated relative to the $2p$ state by orders of

1180: magnitude \citep[see e.g.][and references therein]{D05}. In close

1181: proximity to a luminous source, the $2s$ level is populated mostly via

1182: transitions of the form $1s\overset{{\rm Ly}\beta}{\rightarrow}3p

1183: \overset{{\rm H}\alpha}{\rightarrow}2s$ \citep{Sethi07,FS}, and

1184: $n_{2s}\sim 0.12 n_{1s} P_3/A_{2s1s}$, where $P_3$ is the rate at which

1185: Ly$\beta$ photons are scattered and the prefactor $0.12$ denotes the

1186: probability that absorption of the Ly$\beta$ is followed by

1187: re-emission of an H$\alpha$ photon \citep[][]{FS}.

1188: 

1189: \item The recombination rate into the $2p$ level is given by

1190: $0.68\alpha_{\rm rec,B}n_en_p=1.8\times 10^{-13}(T_{\rm

1191: gas}/10^4\hs{\rm K})^{-0.7}n_en_p$ cm$^{3}$ s$^{-1}$

1192: \citep[e.g.][]{Hui97}, where $n_p$ is the proton density in cm$^{-3}$.

1193: 

1194: \item The rate at which transitions of the form $1s\rightarrow np$

1195: occur by absorbing a photon is given by $P_{n}=4\pi

1196: \int\frac{J(\nu)}{h\nu}\sigma_n(\nu)d\nu$. Assuming for simplicity

1197: that $J(\nu)$ does not vary with frequency, i.e. $J(\nu)=J$, we have

1198: \begin{equation}

1199: P_n=\frac{4\pi J f_n \pi e^2}{h\nu_n m_e c},

1200: \end{equation} 

1201: where $f_n$ denotes the oscillator strength of the transition, $e$

1202: ($m_e$) the charge (mass) of the electron, and $h\nu_n$ denotes the

1203: energy difference between the $1s$ and $np$ levels. The oscillator

1204: strength decreases rapidly with increasing $n$ \citep[e.g. chapter

1205: 10.5 of][]{RL79}, and in practice we can safely ignore all terms with

1206: $n>2$. We then need not worry about the factors $f_{n2}$ \citep[which

1207: have been computed by][]{PF06,Hirata06}.

1208: 

1209: The Ly$\alpha$ scattering rate is given by

1210: \begin{equation}

1211: P_2=\frac{M_F L_{\alpha} f_{2} \pi e^2}{4\pi r^2 \Delta \nu h\nu_{\alpha} m_e c},

1212: \end{equation} 

1213: where we replaced $J$ (in erg s$^{-1}$ Hz$^{-1}$ sr$^{-1}$ cm$^{-2}$)

1214: with $J=M_F\frac{L_\alpha}{16\pi^2 r^2 \Delta \nu}$, in which

1215: $L_{\alpha}$ is the Ly$\alpha$ luminosity of the central source (in

1216: erg s$^{-1}$), and $\Delta \nu$ is the frequency range over which

1217: these Ly$\alpha$ photons have been emitted. The factor $M_F$ takes

1218: into account the fact that resonant scattering traps Ly$\alpha$

1219: photons in an optically thick medium (see

1220: \S~\ref{sec:wind}). Substituting fiducial numbers

1221: \begin{equation}

1222: P_2=2 \times 10^2 \hs {\rm

1223: s}^{-1}\hs\Big{(}\frac{L_{\alpha}}{10^{42}\hs{\rm erg}/{\rm

1224: s}}\Big{)}\Big{(}\frac{10^{-3}\nu_{\alpha}}{\Delta

1225: \nu}\Big{)}\Big{(}\frac{{\rm

1226: pc}}{r}\Big{)}^{2}\Big{(}\frac{M_F}{100}\Big{)}.

1227: \label{eq:lya}

1228: \end{equation} 

1229: 

1230: The rate at which Ly$\beta$ photons scatter can be related to the Ly$\alpha$ luminosity, $L_{\alpha}$, and the equivalent width (EW) of the line, if one writes the specific intensity $J(\nu_{\beta})$ near the Ly$\beta$ resonance in terms of the Ly$\alpha$ luminosity of the central source as $J(\nu_{\beta})=\frac{\lambda_{\beta}}{\nu_{\beta}}\frac{L_{\alpha}}{{\rm EW}}\frac{1}{16\pi^2r^2}$ (note that we assumed that the specific intensity of the continuum remains constant between $\nu_{\alpha}$ and $\nu_{\beta}$). The rate at which Ly$\beta$ photons scatter can then be written as

1231: 

1232: \begin{equation}

1233: P_3=2 \times 10^{-3} \hs {\rm s}^{-1}\hs\Big{(}\frac{L_{\alpha}}{10^{42}\hs{\rm erg}\hs {\rm s}^{-1}}\Big{)} \Big{(}\frac{{\rm EW}}{200\hs \AA}\Big{)}^{-1}\Big{(}\frac{r}{{\rm pc}}\Big{)}^{-2}.

1234: \label{eq:lyb}

1235: \end{equation} 

1236: Equation~(\ref{eq:lyb}) illustrates that it is very difficult to bring

1237: the ratio $n_{2s}/n_{1s}\sim 0.12 P_3/(A_{\rm 2s1s})$ to unity.

1238: 

1239: \item Collisional deexcitation rates relate to the collisional

1240: excitation rates via $C_{lu}=C_{ul}\frac{g_u}{g_l}e^{-\chi/kT}$

1241: \citep[e.g.][]{Osterbrock89}. Combined with the formulas given above, it is

1242: straightforward to verify that the collisional de-excitation rates are

1243: subdominant relative to the rate at which spontaneous Ly$\alpha$

1244: emission de-populates the $2p$ level.

1245: 

1246: \item Lastly, Eq~(\ref{eq:lya}) shows that the stimulated emission rate

1247: is $P_2 \ll A_{21}$.

1248: 

1249: \end{itemize} 

1250: 

1251: \subsection{HI Level Populations in this Paper}

1252: 

1253: The maximum number density of hydrogen nuclei in this paper is

1254: encountered in \S~\ref{sec:wind}, for the expanding shell of HI gas

1255: with $N_{\rm HI}=10^{21}$ cm$^{-2}$ and a thickness $dr=0.01$ kpc, in

1256: which $n_{\rm max}\sim 30$ cm$^{-3}$. At these column densities, the

1257: shell self-shields against ionizing radiation, and is likely mostly

1258: neutral. For simplicity, let us assume that $n_e=n_p=n_{\rm HI}=30$

1259: cm$^{-3}$. Under these conditions:

1260: 

1261: \begin{itemize}

1262: 

1263: \item the collisional excitation rate from $1s\rightarrow 2p$ is

1264: $C_{12} < 3\times 10^{-9}$ s$^{-1}$.

1265: 

1266: \item the collisional excitation rate from $2s\rightarrow 2p$ per atom

1267: in the 1s state- is $C_{2s2p}n_{2s}n_{p}/n_{1s}\sim 5\times

1268: 10^{-2}n_{2s}/n_{1s}$ s$^{-1}$$=5\times 10^{-3}P_3/A_{2s1s}$ s$^{-1}$. The maximum

1269: Ly$\alpha$ luminosity considered in this paper is $\sim 10^{43}$ erg

1270: s$^{-1}$. Let us conservatively assume that EW$=20$ \AA (rest-frame),

1271: which corresponds roughly to the detection threshold that exists in

1272: narrow-band surveys \citep[e.g.][]{Shima06}. Using Eq~\ref{eq:lyb}, we

1273: find that $P_{3,{\rm max}} \sim 10^{-5}$ s$^{-1}$ (for $r=0.1$ kpc), 

1274: and therefore that $C_{2s2p}n_{2s}n_{p}/n_{1s}\sim 10^{-8}$ s$^{-1}$.

1275: 

1276: \item the recombination rate is $5\times 10^{-12}(T_{\rm

1277: gas}/10^4\hs{\rm K})^{-0.7}$ s$^{-1}$.

1278: 

1279: \item the maximum Ly$\alpha$ scattering rate is (substituting $r=0.1$

1280: kpc, $M_F=100$, $\Delta \nu=0.001\nu_{\alpha}$ into Eq~\ref{eq:lya})

1281: $P_{2,{\rm max}}=0.2$ s$^{-1}$.

1282: 

1283: \end{itemize} 

1284: 

1285: By comparing these rates to the rate at which spontaneous emission of

1286: Ly$\alpha$ depopulates the $2p$ state, $A_{21}=6.25\times 10^{8}$

1287: s$^{-1}$, we find that all excitation rates are $\geq 9$ orders of

1288: magnitude smaller than the de-excitation rate for the wind models

1289: discussed in \S~\ref{sec:wind}. In equilibrium, hydrogen atoms in

1290: their electronic ground ($1s$) state are therefore $\geq 9$ orders of

1291: magnitude more abundant than hydrogen atoms in their first excited

1292: ($2p$) state. Furthermore, as was mentioned above the ratio of atoms in the 2s and 1s levels is given by $n_{2s}/n_{1s}=0.12 P_{3,{\rm max}}/A_{2s1s}\sim 10^{-7}$. Since the densities, Ly$\alpha$ luminosities, and the Ly$\beta$ scattering rates are lower, the fraction of HI atoms in their first excited states are even smaller in other sections of the paper. In conclusion, for all applications

1293: presented in this paper, no accuracy is lost by assuming that all

1294: hydrogen atoms occupy their electronic ground state.

1295: 

1296: Finally, in \ref{sec:test} we computed solutions for the radial

1297: dependence of $U(r)$ (Fig~\ref{fig:uden}). The energy density, $U(r)$,

1298: was quoted to depend linearly on the luminosity of the central

1299: source. This is valid unless ({\it i}) the Ly$\alpha$ scattering rate, $P_2> A_{21}=6.25\times 10^8$ s$^{-1}$, or ({\it ii}) the Ly$\beta$ scattering rate exceeds $P_{\rm 3} \gsim 10^2$ s$^{-1}$.  In either case, our

1300: assumption that all hydrogen atoms populate their electronic ground

1301: state breaks down. Condition ({\it i}) translates to $L_{\alpha}\gsim 3\times 10^{48}$ erg s$^{-1}(\Delta \nu/0.001\nu_{\alpha})(r/{\rm pc})^2(100/M_F)$ (Eq.~\ref{eq:lya}), while condition ({\it ii}) translates to $L_{\alpha}\gsim 10^{46}$ erg s$^{-1}({\rm EW}/20\hs{\rm \AA})(r/{\rm

1302: pc})^2$ (Eq.~\ref{eq:lyb}). Substituting $r=0.1$ kpc, $M_F=100$ (Fig~\ref{fig:uden} shows that resonant scattering enhances the energy density by a factor of $\gsim 10$ relative to $L_{\alpha}/4\pi r^2 c$), $\Delta \nu=10^{-3}\nu_{\alpha}$

1303: (thermal broadening alone in $T=10^4$ K gas results in $\Delta v\sim 26$ km s$^{-1}$, which translates to $\Delta \nu=10^{-4}\nu_{\alpha}$), and $EW=20$\AA, condition ({\it i}) translates to $L_{\alpha}\gsim 3\times 10^{52}$ erg s$^{-1}$, while condition ({\it ii}) translates to $L_{\alpha}\gsim 10^{50}$ erg s$^{-1}$

1304:  The more conservative condition ({\it ii}), $L_{\alpha}\gsim 10^{50}$ erg s$^{-1}$, translates to $\dot{N}_{54} \gsim 10^7$, well beyond the regime considered in this paper.

1305: \label{lastpage}

1306: \end{document}

1307: